negative RSquare for test data, while train data RSquare is good

Question

I am new to linear model , train_test_split in sklearn.model_selection

I follow a standard "format/procedure" to train a model for an input CSV file related